clover carmen wrote: (original post HERE)
i wanted to touch on a few points you've made with some remarks and rebuttals. not at all meant as an attack, but instead a continuance of the conversation.
The television and calculator faced similar backlash, but we have never experienced a method of learning and thinking that completely replaces your reception, reaction, and solution all in one. For higher level questions in mathematics, the calculation is not the hard part, understanding what is being asked is. There's a reason why they allow calculators for the math portion of the SAT. When an LLM can easily replace all brain power in solutions, there is a missed opportunity to develop your own methods and heuristics.
There is a theoretical version of AI which I would accept. If the AI learned to draw. If an AI was given a stylus and over the course of several thousands of iterations was tested on its ability to draw a bird, I could see that being an AI's own art. It developed a skill over time which involved feedback and ingenuity.
(this is called deep reinforcement learning and is best explained through example, one of my favorite being this video on AI learning to walk)
AI generated images do not do this. there is no emergent behavior in these generators. it is at best, a prediction based on what it was fed. If you feed it thousands of real life photographs of a bird, it will never be able to draw a bird. Cavemen saw birds and were somehow able to draw a bird, but a neural network will never be able to create for itself. Not because it's stupid, but because that's not in its realm of capabilities. It can only recycle ideas, it can never create.
Inspiration requires innovation, generative AI image generators are incapable of such.
and this does not at all touch on the environmental impacts of AI, which while are fundamentally important to this conversation, are theoretically fixable using new sources of energy and an emphasis on ethics in its development.
really really tried to come at this with some meaningful and reasonable arguments. if we have a fundamental disagreement on certain subjects, i understand and will not devolve this into bickering. i hope any of this makes sense, i'm hella sick rn. really hope you're well and hope to see more of you around <3
I’m glad someone finally opened this topic, because I have a lot to say about AI. As an artist, I don’t mind AI itself. It’s just a tool that can help or harm depending on how it’s used. I use ChatGPT as a secretary, brainstorming buddy, and...hello beloved shop designer and fashionable ximbo clover carmen :) been a hot sec since i've seen you around!
i wanted to touch on a few points you've made with some remarks and rebuttals. not at all meant as an attack, but instead a continuance of the conversation.
AI should serve us, like in Star Trek or The Jetsons, freeing humans from drudgery and giving us access to knowledge, medicine, food, and housing for free so we don't have to work anymore to earn the things that should be given to us for free since they are basic human rights like houses, food, medicine, clothes, knowledge...etc.i agree that the scientific uses of AI have me very hopeful. The AI-assisted simulations made for medical research greatly accelerate the process of diagnostics, discovery, and treatment. my senior design class this year had a multitude of medical AI projects, which required an incredible amount of work and provided excellent results. one of these projects was to diagnose pediatric pancreatitis and breast cancer using data sets provided by a local hospital, and it achieved an excellent accuracy rate in diagnosis. another was to analyze CT scans and allow users to compare their own scans with that of diagnosed patients with cardiac illnesses. it was entirely seen over by the user's medical practitioner and sought to give educated guesses and comparisons for users to better understand their results. and that's just the work made by college students! incredible steps have been made by industry professionals that offer phenomenal results and hope for the future of medicine.
There have been many papers on this subject, but i'll link to one [Generative AI Can Harm Learning by Wharton University of Pennsylvania]. Repetitive use of AI in place of traditional learning significantly decreases a student's performance in controlled testing environments " In stark contrast, in the subsequent unassisted exam, student performance in the GPT Base cohort degraded by 0.054 (out of 1) relative to that of the control cohort. In other words, GPT Base diminished the average control student’s performance on the unassisted exam by 17%" There is a developing over-reliance on LLMs such as ChatGPT for easy answers in all subjects. Gone are the days where cheaters had to cross-check Chegg with Brainly and solution documents on the internet to ascertain what answer made the best sense to them. Gone are the days where students had to examine their study materials and come up with key points and figures on their own based on what their mind decided was of the most importance and most unfamiliar to them. Instead, students can now upload their homework PDF and have the work instantly displayed, explained, and have work shown. Most of these students do not double check, let alone read these answers before submitting them. The amount of times my professors have received papers that start with "Awesome! Here's your paper..." because it was simply copy and pasted. Cheating students no longer need to go through and paraphrase existing papers (by proxy actually reading them) to bypass plagiarism checkers. AI has a distinct writing style, yes. But not only have professors who run papers through checkers faced intense backlash, they're also incredibly inaccurate in their current state. Learning and reading your materials, cheated or not, builds neural pathways in your brain that allow for you to learn and think for yourself. Copy and pasting ChatGPT answers degrades your learning and leaves students feeling dependent and incapable of developing their own thoughts and solutions. In a world where you needn't think, why would you.Anti‑AI claims I reject:
“AI replaces critical thinking.” Wrong. They said the same thing when they invented the TV: it depends on how you use it.
The television and calculator faced similar backlash, but we have never experienced a method of learning and thinking that completely replaces your reception, reaction, and solution all in one. For higher level questions in mathematics, the calculation is not the hard part, understanding what is being asked is. There's a reason why they allow calculators for the math portion of the SAT. When an LLM can easily replace all brain power in solutions, there is a missed opportunity to develop your own methods and heuristics.
“AI steals art.” An AI‑bro countered: “Steal from one source and you’re a thief; steal from many and it’s inspiration.” That left me speechless. If someone knows how to counter this argument, please reply below.I hear this a lot. And yes, it is initially stumping for many. To be inspired is to consume artwork for many styles and develop something entirely your own. There is a process of incorporating methods and techniques from other artists, that you then implement in your own way. AI does not do this. There is no personal flair or innovation by an AI. If a human is shown two photographs of a bird, it could draw a bird. If an AI is shown two photographs of a bird, it will generate static.
There is a theoretical version of AI which I would accept. If the AI learned to draw. If an AI was given a stylus and over the course of several thousands of iterations was tested on its ability to draw a bird, I could see that being an AI's own art. It developed a skill over time which involved feedback and ingenuity.
(this is called deep reinforcement learning and is best explained through example, one of my favorite being this video on AI learning to walk)
AI generated images do not do this. there is no emergent behavior in these generators. it is at best, a prediction based on what it was fed. If you feed it thousands of real life photographs of a bird, it will never be able to draw a bird. Cavemen saw birds and were somehow able to draw a bird, but a neural network will never be able to create for itself. Not because it's stupid, but because that's not in its realm of capabilities. It can only recycle ideas, it can never create.
Inspiration requires innovation, generative AI image generators are incapable of such.
“AI will never replace creativity.” It already produces work indistinguishable from real footage. Give it some time and it will create "creativity".Creativity is exclusive to those with real life experiences and emotions. it is inherent and natural. it will never fully go away. however, the environment for creatives grows harsher and harsher. Animation giant Disney who while many disagree with, you could at least say they were staunchly against AI replacing creatives, has betrayed their artists by announcing their introduction of generative AI into their platforms. The appreciation we once had for human artwork has been min-maxxed once again to an extent to where, quite frankly, we're replacable to most industries. AI will not replace creativity in humans, but it will replace creative work in its most important industries. Again, the way that generative AI works does not allow for creativity. It will never learn to draw a bird if only shown photographs.
and this does not at all touch on the environmental impacts of AI, which while are fundamentally important to this conversation, are theoretically fixable using new sources of energy and an emphasis on ethics in its development.
really really tried to come at this with some meaningful and reasonable arguments. if we have a fundamental disagreement on certain subjects, i understand and will not devolve this into bickering. i hope any of this makes sense, i'm hella sick rn. really hope you're well and hope to see more of you around <3
